IIRC the official reason was that some automated anti-spam code accidentally caught the kbin user agent and mistakenly added it to a block list, and the lemmy.ml admins were busy and didn't see it for over a week - but once one of them noticed it was promptly fixed.

Also, I recall this being specific to lemmy.ml - other instances run by other admins like lemmy.world and lemmy.ca weren't affected.

Unfortunately I don't know. It could very well have been that their own instances were being flooded by traffic from kbin.social and this was their solution. However, to my understanding this was added to the source code itself so all lemmy instances upgrading suddenly become an issue for kbin instances. I had to change the user agent in my own kbin instance to stop the flood of errors as it was absolutely messing up the.coolest.zone and I basically couldn't troubleshoot any legitimate failed messages.

Whatever the case is, I'm very glad it's resolved now.